ich habe jetzt ein bisschen mit KI rumprobiert. Im PP Modus hatte sich der Motor mal gedreht im CSP Modus nicht mehr. Scheinbar hatte irgendwie die DC Synchronisation gefehlt. Hier mein aktueller Stand:Wir haben bereits bewiesen:

• EtherCAT OP ✅
• DC-Synchronisation 1 ms ✅
• CSP PDO-Mapping ✅
• Code:

  0x607A
  über PDO erreichbar ✅
• Motor reagiert auf kleine CSP-Positionsänderungen ✅
• aktueller Fehler
  Code:

  0x603F = 0
  ✅

ich würde jetzt gerne G0 X10 in LinuxCNC eingeben und gucken ob der Motor sich dreht, aber ich kann keine Referenzfahrt durchführen. Und ohne kommt jedes mal die Meldung Bewegung würde pos Anschlag von Achse überschreiten.
 

Take the time to read the cia402.comp source and example. I was wrong, cia492.comp includes a pin to select PV (velocity) mode which is what you want. 

I'd probably edit the name in your XML to be S0 instead of D1. this will make more sense when assiging pins to spindle.0
There are pins for target_velocity and actual_velocity (encoder feedback) so you are getting close. I think that your spindle speed-cmd pin should map to target_velocity and be sure the encoder is properly scaled.
I'm not sure how to map spindle.0.on to enable the motor.
spindle at speed would use the near component between  target_velocity and actual_velocity
to capture a long list of pins you can add > pins.txt to the show pin command to write output to a file

JT is expressing a view some developers have. Personally, I find there are so many simulations now the alternative is downright confusing  when looking for a specific sim. I think the solution is refactoring our git repo to support additional guis in a plugin type arrangement. That is not a trivial task.

20 or 25 MHz (2.4x or 3x faster than the default 8.3333 MHz), works fine with no skew added.
I would be comfortable with those settings.

The frequency value you set will be rounded to the closest available frequency
= an integer division of the 100 Mhz FPGA ClockLow. This is easy to test with halshow.

As you go faster, the timing margins become smaller and you must adjust to the delay in
the actual hardware multiplexing with the skew adjust (in ns) but at the higher speeds
the timing variations over voltage and temperature limit the practical maximum.

That is, I can get all encoders to work with a 50 MHz mux rate (and 20 ns skew) but I would not trust
it without doing at least a temperature cycle.

Timing errors will show up as even/odd encoder swaps or crossover between even/odd encoders
so don't experiment on a powered servo system...

tuxcnc wrote:

rodw wrote:

I don't like using backports. Never have used them.
 
"I don't use it" and "I don't like it" are not substantive arguments.
The problem is that there are newer versions of the libraries in the newer release. First of all, it is about Python, in which the authors constantly change or remove something, so even relatively new programs (from a year or two ago) do not want to run, or some of their functions do not work.
However, backports repositories are optional, you can activate them, install only what you need and then turn them off again.

The reason I don't like backports is it can land you in dependency hell as you point out. That's my substantive argument.
I've been in this position needing a later kernel to support my hardware. Today, I would either compile the kernel and be done with it or use apt pinning to be very selective on what packages I obtain from unstable.

Hello everyone I have a new issue with my plasma table. So after I home my machine, load the file, zero the machine, and click cycle start my z axis torch travels up into the home switch instead of going down and probing the sheet of steel. I ran an mdi comand "g0 z-.5 and when I hit enter the torch moved down -.5 like it should so I dont think anything is reversed. The gcode file I am using worked in the past so im am just at a loss. Is there a certain amount of travel the torch is supposed to go up initially that I could maybe adjust to fix this? 

 
 

You could use one of the Mesa 7i76e's not very accurate analog voltage input pins and a 10k resistor to form a voltage divider. at 25 deg C, and 24 volt field power, the voltge will be 12 volts. You might need to scale this with lincurve or a custom component. Accuracy of +- 2 degrees due to 8 bit conversion is probably OK for this application.

I have a Jianken 3.2kw spindle that has a NTC 103F3950 (10 kΩ, B=3950) thermistor and a Thermal fuse / PTC-type cutoff.
My controllers are Mesa 7i76e + 7i78 + 7i89.
How do I make use of these and connect them to my controllers.

Oh, yes I understand,
Thx

Gotrune wrote:

Code:

[code][code]my high-resolution optical scales (...) are dropping pulses during high acceleration

[/code][/code]

Acceleration has nothing to do with it.
Only maximum speed matters.
From the resolution and speed you will calculate the pulse frequency.

Hi PCW,

Thank you so much for the incredibly fast reply!

I will add setp hm2_7i97.0.encoder.muxed-sample-frequency 20000000 to my HAL file right away and test the machine.

Just to make sure I fully understand the limits of the hardware: what are the acceptable or maximum values for this parameter with the standard 7i97T bitfile? For example, does it need to be a specific integer divisor of the base clock? Also, at what frequency would I typically need to start adjusting the muxed-skew parameter you mentioned?

Thanks again for your time and excellent support!

I would try setting:

hm2_7i97.0.encoder.muxed-sample-frequency

to say 20 MHz

(you can probably go higher by futzing with the muxed- skew parameter)
 

mesaboard is on 2. enp8s0 (see alsö static inet/inet6 addresses) It does show no carrier when unplugged.

eno1 (internet) has automatic connection method.

wlan has been turned off in order not to get confused.

I changed lan cables and now it behaves as it should mesact/machine finds the board and readhmid provides the information as expected.(mesact and terminal)
I wasnt aware of that since I did have acces to the internet anyhow(???).

so thank you very much for the moment This particular problem is solved.


Have nice days

geebee